Fractures of the hip and wrist are a costly and preventable threat to the health and survival of the elderly. A better understanding of the risk factors for hip and Colles' fractures will add to our knowledge about the etiology of this condition and help focus research and preventive efforts on those who are at greatest risk. We propose to characterize risk factors for hip and Colles' fractures in a multicenter prospective study of 9,600 elderly women followed quarterly for an average of 3.5 years. Our cohort will be a population-based sample of non-black women aged 65 or greater who are screened from 1986 thru 1988 at clinical centers in Pittsburgh, Portland, Maryland and Minneapolis. The Portland Clinical Center will attempt to recruit its entire sample from the membership of Kaiser Permanente, Northwest Region. The Center for Health Research has access to all medical of the KP membership and has extensive experience in recruitment and follow-up of large populations. At the baseline exam, we will measure a large set of potential historical and clinical predictors of fractures. We will use single-photon absorptiometry of the radius and also x-ray the hand, spine and hips to determine whether simple radiographic measurements can identify those with the greatest risk of fracture. We will also freeze serum from all participants at the outset and test the stability of hormone measurements on a subsample after storage at -196 degrees C for several years; measurements that are shown to be stable will be the subject of a future study of the hormonal differences between women who suffer fractures and those who do not. We conservatively estimate that 165 hip and 150 Colles' fractures will occur, yielding a power of .90 to detect risk ratios of less than 2.0 for common risk factors. Our prospective design will yield more information with fewer biases than the previous (retrospective) studies on this topic, and our nested-case control analyses will minimize the costs of the expensive endocrinologic and radiologic measurements. The study cohort, and its bank of sera, x-rays and clinical data is an excellent resource for future studies of osteoporosis, fractures and other diseases of the elderly. The results of our study will guide future research and rationalize clinical and public health efforts to prevent these fractures.